Python Graph API

This is a minimal Python API for the Graph. It works with Python 2 as well as Python 3.

Installation

Optionally, create and activate a virtual environment, so that installations are not system-wide:

$ python -m venv <path to venv>
$ source <path to venv>/bin/activate

The Python Graph API can be installed with pip directly from our git repository:

$ pip install git+git://git.graph-it.com/graphit/graph-client-py.git

The Graph API uses msgpack. So, the Python binding for it will be installed as a dependency.

In order to use TLS connections to graphs, a client certificate chain including the private key has to be available in a file. (Internally, SSLContext.load_cert_chain() is used to establish the TLS connection.)

In our current graph installations a chain file with a private key can be found in /var/lib/graph/var/cert/srv/local.srv.pem:

$ scp graph.example.com:/var/lib/graph/var/cert/srv/local.srv.pem graph.crt

Usage

A graph connection is initialised by giving a URL and (if it is a TLS URL) the filename of the certificate chain to the constructor:

>>> import graph
>>> gc = graph.Connection('tls://graph.example.com:4439', 'graph.crt')

Then all functions of the Graph API can be called on the connection object:

>>> graphmodul_guid = gc.attributsknoten('graphmodul_name', 'graph')
>>> knotens = gc.listeattribute(graphmodul_guid, 'knoten', 'typ, dokumentation, anzahlattribute', 'typ asc')

Observe that all string data returned by the API are bytestrings (even the keys in *attribute results) and have to be decoded to be used as Python strings:

>>> for knoten in knotens.values():
...     print("{}: {} ({})".format(knoten[b'typ'].decode(), knoten[b'dokumentation'].decode(), knoten[b'anzahlattribute'].decode()))
... 
aktion: <p>Auszuf&uuml;hrende Aktion</p>
 (5)
aktionfunktion: <p>Auszuführende Funktion, die an einem Knoten hinterlegt werden und von Änderung an lokalen Attributen getriggert werden kann.</p>
 (22)
attribut: <p>Allgemeines Merkmal</p>
 (16)
[…]

Test script

The test.py script reads the configuration of a main graph from config.json, queries this main graph for remote graphs configured in it, prints a summary of all graph modules in the main graph and all remote graphs and downloads stationeries in all the graphs if they are configured there.

Plans

Up to now, only a raw connection is implemented, i.e., the Python implementation does not know anything about the actual API calls and just passes them to the graph server.

We should add implementation and documentation for these actual API calls. This would also allow to do at least some of the necessary conversions already in the library.